Hi, we’re having trouble getting our Arducam B0346 (IMX577) camera module to output any image on a Qualcomm development board. The OV9282 works fine on the same board, but the IMX577 just shows a black screen / no preview. The module is connected through a CRI adapter board to the CSI connector on a CRI Qualcomm reference board. One odd thing — the box is printed IMX477, but the module is supposed to be an IMX577. Here are the logs we captured:
[ 1513.141010] CAM_INFO: CAM-SENSOR: cam_sensor_driver_cmd: 977 Probe success,slot:1,slave_addr:0x34,sensor_id:0x577, is always on: 0
[ 1528.128670] CAM_ERR: CAM-ISP: cam_ife_csid_irq: 5014 CSID:1 RX_ERROR_LANE0_FIFO_OVERFLOW: Skew/Less Data on lanes/ Slow csid clock:400000000Hz
Thanks for the logs. The first line is actually good news — the sensor probes successfully (sensor_id 0x577), so I2C communication and sensor ID detection are working. The second line is where it fails: RX_ERROR_LANE0_FIFO_OVERFLOW on CSID:1 usually points to a MIPI lane data/clock issue rather than an init problem. Can you share your current sensor driver/parameter config and the camera initialization code? Also, what platform is this on — Android or Linux?
We’re on Qualcomm, and we’ve tested both Android and Ubuntu. We already tried the configuration file your support provided earlier, but still no image. I’ve attached our parameter file and the full log for reference. The initialization finds the sensor fine — it’s only when we start preview that it fails.
Got it. Since the probe succeeds but the CSID reports FIFO overflow on lane 0, the sensor is answering on I2C but the MIPI data path isn’t delivering valid data to the CSID. ‘Skew/Less Data on lanes’ can mean the lane count, data rate, or clock settings don’t match what the sensor is actually outputting. Have you tried a different FFC cable length? Sometimes a long or marginal cable causes exactly this. Also, does swapping the IMX577 for the OV9282 on the exact same adapter + cable + connector give a normal image?
We replaced the FFC with a 60mm one and still get no output. And yes, if I swap in the OV9282 on the same CRI adapter and Qualcomm setup, the image comes up fine. So the board/adapter side looks okay.
That’s a useful data point. Since the OV9282 works on the same path, the board, adapter, and CSI connector are probably fine, and the difference is in how the IMX577 is configured versus the OV9282 (lane count, data rate, clock, etc.) or the IMX577 module’s MIPI output itself. I’ve put together a different config file for the IMX577 and sent it to you via private message — please try it and post back what you get. The FIFO overflow usually means the CSID clock (400 MHz in your log) and the incoming lane data aren’t agreeing.
Tried the new config file, still black screen with the same CSID error. Question — has this IMX577 parameter set actually been validated on a Qualcomm platform? We’ve tried both Qualcomm Android and Ubuntu and neither shows a preview.
Honestly, we have not run this exact IMX577 config on a Qualcomm platform, and we don’t have your hardware set here, so we can’t reproduce it on our side right now. The next step should be to verify the camera’s MIPI output first, before chasing platform config. Can you scope the MIPI lane signals (lane 0 at least, plus the clock lane) while streaming? With the config file I sent, check whether the sensor is actually driving lanes with data and whether the clock is present. If the sensor isn’t outputting valid MIPI data, no platform-side setting will fix it.
Understood. While we’re checking that — do you have a 4-lane version of the IMX577, or any other IMX 12MP+ modules that might work better here? One thing we noticed is the OV9282 only started working on the Qualcomm board after we did a DCDC conversion on its power.
To summarize where we are: the IMX577 (B0346) is detected over I2C (sensor_id 0x577), but preview fails with CSID lane0 FIFO overflow, which points to a MIPI data/clock mismatch or a module output issue. Since the OV9282 works on the same adapter/board, the platform CSI path is likely fine. We haven’t validated the IMX577 config on Qualcomm and can’t reproduce it without your hardware, so please measure the MIPI lanes with the config I sent to confirm the sensor is actually driving valid data — that determines whether this is a module/signal problem or a platform config problem. For module options (4-lane IMX577 or other 12MP+ IMX variants), I’ll send you a few relevant product details in a private message so this thread stays on the debugging track. Let me know what you see on the scope.